Rapid imaging and machine learning-based evaluation of drug sensitivity in primary cells

We are developing a label-free imaging platform to enable rapid, high-throughput genetic and pharmacological screening of primary cancer cells, including non-dividing ones, without the need for intermediate model generation. By using malignant ascites from gastroesophageal cancer patients, we are optimizing workflows to preserve viability and heterogeneity within 24 hours of collection. Predictive models, trained on brightfield images, achieve high accuracy in inferring cell identity and viability. Our approach expands the range of patient samples that can be studied, enhancing preclinical discovery efforts and paving the way for personalized therapy nomination.

biosensor-image

Principal Investigators

Jesse Boehm, PhD (MIT)

Samuel Klempner, MD (MGH)

Beth Cimini, PhD (Broad Institute)

Koch InstituteBroad InstituteMGH
Experiment:
AscitesPredict
Experiment:
Experiment:

Principal Investigators

Jesse Boehm, PhD (MIT)

Samuel Klempner, MD (MGH)

Beth Cimini, PhD (Broad Institute)

Koch InstituteBroad InstituteMGH
HTS
Street Address
500 Main St
Bldg 76, Rm 043
Cambridge, MA 02139
Shipping Address
Attn: HTS (76-043)
32 Vassar St
Cambridge, MA 02139
hts-lab@mit.edu
Koch Institute